In-network live stream processing with named functions

Abstract

Information Centric Networking (ICN) is designed for content distribution. Therefore, it is a solid basis for data streaming applications. In general, data streaming applications have become very popular in networks where many sensors provide their data. Thereby, users often want to fetch the information of several data streams and aggregate or analyze them. Normally, such results are smaller than the input data. Therefore, it can be beneficial to compute the analysis on a data source. Especially for sensor devices this may not be possible due to their limited computation power. Additionally, sensor networks can be widely distributed, meaning that to compute the analysis on one sensor device requires to aggregate all streaming data on it. Hence, it makes sense to aggregate or analyze the streaming data inside the network. Here Named Function Networking (NFN) can help. It is an ICN generalization which enables users to request results of in-network computations. We show how user defined in-network data analysis can look like in several scenarios such as analysis of live streaming GPS tracking data. Thereby, we present how the NFN naming scheme can be used to name data streams. Furthermore, we address the problem how to enter an existing data stream.